Mice Expressing RHAG and RHD Human Blood Group Genes

Anti-RhD prophylaxis of haemolytic disease of the fetus and newborn (HDFN) is highly effective, but as the suppressive mechanism remains uncertain, a mouse model would be of interest. Here we have generated transgenic mice expressing human RhAG and RhD erythrocyte membrane proteins in the presence and, for human RhAG, in the absence, of mouse Rhag. Human RhAG associates with mouse Rh but not mouse Rhag on red blood cells. In Rhag knockout mice transgenic for human RHAG, the mouse Rh protein is “rescued” (re-expressed), and co-immunoprecipitates with human RhAG, indicating the presence of hetero-complexes which associate mouse and human proteins. RhD antigen was expressed from a human RHD gene on a BAC or from RHD cDNA under control of β-globin regulatory elements. RhD was never observed alone, strongly indicative that its expression absolutely depends on the presence of transgenic human RhAG. This first expression of RhD in mice is an important step in the creation of a mouse model of RhD allo-immunisation and HDFN, in conjunction with the Rh-Rhag knockout mice we have developed previously.


Introduction
The human Rh (Rhesus) blood group is of clinical interest due to its role in transfusion medicine, auto-immune anaemia and its implication in materno-fetal incompatibility and hemolytic disease of the fetus and newborn (HDFN) [1]. RhD is a highly immunogenic antigen and despite the effectiveness of RhD prophylaxis, materno-fetal immunisation due to RhD antigen, a cause of perinatal mortality and morbidity, is not completely eradicated [2][3][4]. Currently, monoclonal or recombinant anti-D are being tested for prophylaxis as alternatives to plasma derived polyclonal IgG anti-D. Although a number of models have been developed, the mechanism for anti-RhD suppression has as yet to be determined (reviewed in [5,6]). In this context, a transgenic mouse model of RhD antigen expression would be useful to study RhD alloimmunisation and HDFN.
Rh antigens are present in the erythrocyte membrane in an oligomeric association of two major components, Rh proteins (RhD and/or RhCcEe), and homologous Rh-associated glycoprotein (RhAG). The Rh complex also includes other proteins (ICAM-4/LW, CD47/IAP and glycophorin B), all linked by non-covalent bonds (reviewed in [7][8][9][10]). Rh-deficiency, a rare autosomal recessive disorder in man, is caused by mutations occurring either in the RHAG or RH locus. When the RhAG or the Rh subunit is absent, the Rh complex is missing or severely reduced [8,11].
Rh and RhAG proteins, which compose the core of the Rh complex, interact together within what is most likely to be a trimeric structure, based on crystal structure of RhCG [12], NeRh50 [13] and AmtB [14,15]. This is supported by transmission electron microscopy of the human homolog RhCG expressed in, and purified from, HEK293E cells [16]. A predictive model integrating the Rh-RhAG core complex as hetero-trimer within the AE1 multi-protein complex linking the membrane to the cytoskeleton has been proposed [17,18]. In mouse, the relationship between mRh and mRhag as well as other members of the Rh complex, differs somewhat from that in human erythrocytes. Notably, mRhag glycoprotein is less dependent on Rh in the mouse than in man: in knockout mice, mRhag is expressed in the absence of mRh, though at slightly reduced levels, but mRh cannot be expressed without mRhag [19]. CD47, accessory protein of the Rh complex in man, is independent of the complex in the mouse: CD47 is severely defective in 4.2-deficiency in man but not in mice [20,21], mouse CD47 lacks the cytoskeletal connectivity of the human protein [22] and Rh or Rhag knockout mice have no defect of mCD47 [19]. ICAM4 is associated with the Rh antigens and absent in Rh null phenotypes in mouse [19] as well as in man.
We have developed two transgenesis approaches to express the RhD antigen, using human genomic DNA in a bacterial artificial chromosome, and a β-globin LCR-promoter system [23]. It was established, both in vitro in mouse erythroleukemia MEL-C88 cell line, and in vivo, that mouse mRhag is not sufficient to allow expression of human RhD, which must be partnered with human RhAG (hRhAG) to reach the erythrocyte membrane. For the first time, mice expressing human RhD antigen on their erythrocyte membrane have been produced, and the various transgenic mice obtained provided new information on the Rh complex.

pGSEL1 vectors, MEL-C88 cell line culture and transfection
Murine erythroleukemia line MEL-C88 [25] and pGSEL1 vector system [23] were a gift from M. Hollis (then at ICI Pharmaceuticals, UK). Human RHAG, RHD and RHce cDNAs ( Figure 1A) were cloned into this system allowing protein expression under control of the β-globin LCR/promoter/ enhancer.

Transgenic mice
BAC selection and injection. BACs containing RHD and RHAG genes were identified by PCR screening of the bank of human BACs in pBELOBAC11 at the Centre d'Etude du Polymorphisme Humain (CEPH, Paris). Four BAC clones positive for the RHD gene and 8 for the RHAG gene were identified. These were further characterized for size and extension by pulse field gel electrophoresis (PFGE) and by PCR to ascertain the presence of complete gene and, for RHD, absence of RHCE. For injection, a BAC (CEPH n°H193H9) of the RHD group and a BAC (CEPH n°H0696H07) of the RHAG were selected (Table S1).
Transgenesis experiments were carried out at the Institut Pasteur, in the Laboratoire de Biologie du Développement (Pr Ch. Babinet), later at the Centre d'Ingénierie Génétique Murine (Dr F. Langa). BACs were micro-injected in circular form into fertilized C57Black6/SJL mouse oocytes.
Transgenesis of hRHAG and RHD under control of the βglobin LCR/promoter/enhancer was carried out at the Service d'Expérimentation Animale de Transgénèse (SEAT) / UPS44 CNRS Villejuif (Drs.Goujet-Zalc and Martin). RHAG and RHD transgenes under control of the erythroid-specific β-globin promoter and LCR, excised from pEC3-pGSEL1 with AatII to eliminate vector sequence, were purified and micro-injected into C57Bl/6 and B6/CBA fertilised oocytes.
Mouse breeding (including with Rhag -/-mice [19]) and blood collection were carried out at SEAT /UPS44 CNRS Villejuif and at TAAM-CDTA /UPS44 CNRS (Orléans). All work was done in accordance with Policies and Directives for the Care and Use of Laboratory Animals of the CNRS, in compliance with French and European Union animal welfare policies. Blood samples were transferred to INTS for analysis. Personal licenses from the French Veterinary Services were N°75-862 for JPC, 75-1236 for DG, A 92-368 for SM.

Flow cytometry analysis
Surface antigen expression on cells or erythrocytes was determined by immunostaining, followed by flow cytometric analysis on a BD-FACSCanto™ II flow cytometer (Becton-Dickinson). When appropriate, cell viabilility was determined using a viability indicator such as TO-PRO®-1 iodide from Molecular Probes (abbreviated TOPRO1). Quantitative determination of surface antigens (antibody binding capacity) was done with QIFIKIT (DAKO). For intracellular epitope analysis (mRhag, mRh), erythrocytes were fixed in formaldehyde-glutaraldehyde, permeabilised with octyl- glucoside, blocked, immunostained and analysed by flow cytometry as described [26].

Quantitative Real-time PCR
Total RNA was isolated from whole blood of WT , BAC1 or BAC1 x 68.08 transgenic mice, using the High Pure RNA Isolation kit (Roche). No lysis of erythrocytes was performed to ensure reticulocytes were retained. cDNA was synthesized by reverse transcription-PCR (RT-PCR) using the SuperScript® VILO™ cDNA Synthesis Kit (Life Technologies).
Quantitative real-time PCR was performed on the Applied Biosystems 7300 real-time PCR system and TaqMan® Endogenous Control mouse transferrin receptor-Mm00441941 (noted mTfR1) in duplex with TaqMan® gene expression assays (Life Technologies), with exon-spanning probes, for endogenous mouse Rh (Rhd mouse Mm00456910-m1), or transgenic human RhD (RhD/CE human Hs00414315-m1). Assay specificity was verified with the following negative controls: non-transfected mouse samples for RHD and mRh knockout samples for mouse Rh. Amplification efficiencies (E) were comparable for the three primer pairs (1.96-2). All samples were measured in triplicate, and two reverse transcriptions were tested. Standard curves for control and genes of interest were included in each experiment.No template controls and No RT controls were included in each assay, and amplification and detection were performed under standard conditions as recommended by manufacturer. The fluorescent signal intensities were recorded and analysed during PCR amplification using the Sequence Detection Software (SDS) software (Applied Biosystems). Efficiency corrected relative expression ratios for gene of interest (GOI) to mTfR1 were calculated as R=(E target ) ΔCp target (control-sample )/ (E reference ) ΔCp reference (control-sample) , with pooled cDNA from GOI positive mice as controls.
Stopped-flow assays. Stopped-flow experiments were performed at 15°C, as previously described [28]. Resealed ghost pellets were diluted at 1-2% cytocrit in the incubation medium (50 mM K 2 SO 4 10 mM Hepes/KOH, pH 7.2) and mixed (vol/vol) with ammonium buffer (40 mM K 2 SO 4 10mM (NH 4 ) 2 SO 4 10 mM Hepes/KOH, pH 7.2) in the stopped-flow instrument (SFM400, Bio-Logic, Grenoble, France), generating an inwardly directed 10 meq NH 4 + gradient. Data from six to eight time-courses were averaged and fitted to monoexponential function using the simplex procedure of the Biokine software (Bio-logic). The excitation wavelength was 465 nm and emitted light was filtered with a 520 nm cut-off filter. The intracellular pH-dependent fluorescence changes were followed and analysed, as fluorescence increase corresponded to pH elevation. Over the pH range used (6.8-7.8) the relative fluorescence of the dye was proportional to pH. Kinetic rate constants (k) were compared as different values of k correspond to different ghost permeabilities [28]. From the deduced alkalinisation rate constants (k), the apparent NH 3 permeability was calculated following the equation P' NH3 =k exp . (V 0 / SA), with V 0 / SA the ghost-volume/surface-area ratio. Unitary permeability was calculated as Punit NH 3 = P' NH3 hRhAG x1/ N/SA with N/SA the number of sites per surface unit. Ghost diameter measurements were performed using an Axio Observer Z1 microscope (equipped with an AxioCam MRm camera) as described [29].

hRhAG, RhD and Rhc membrane protein expression in MEL-C88 erythroleukemia cell line
Expression of human hRhAG, RhD and Rhc proteins in a murine context was assessed using cDNA constructs under control of the EF1α promoter. In MEL-C88, hRhAG protein could be expressed alone as seen by flow cytometry, and stable cell lines established ( Figure 1A, left). With the β-globin LCR/promoter/enhancer ( Figure 1A, right), hRhAG was expressed, at a lower level. This model also demonstrated that human RhD and Rhc antigens reached the cell membrane only when co-expressed with hRhAG ( Figure 1B). In both cases erythroid differentiation by DMSO treatment increased expression in transient transfection. Human RhD expression was dose-dependent on hRhAG expression in MEL-C88 cotransfected by hRHAG and RHD ( Figure 1C).
These experiments showed that in presence of hRhAG, human RhD and Rhc antigens could be expressed in a murine context, whereas MEL-C88 cells transfected with the human RHD transgene alone do not express the RhD protein.

RHD transcript but no protein expression in mice transgenic for RHD_ BAC1
Transgenesis was carried out using hRHAG and RHD genomic constructs from a human BAC library. No transgenic founder could be obtained for hRHAG; one was identified for RHD (TG_RHD_BAC1) ( Figure S1B, Southern blot) and crossed with a C57Black6/SJL male. Erythrocytes of transgene-positive F1 mice did not react with a panel of anti-RhD monoclonal antibodies recognizing different epitopes or a polyclonal anti-RhD (data not shown). Crossing F1 hemizygotes yielded 6/9 pups carrying the RHD transgene, but none expressed RhD, as shown by flow cytometry 3). Realtime RT-PCR on F2 blood samples yielded specific amplification, indicating presence of RHD mRNA, and cloning and sequencing of the RHD cDNA from bone marrow showed a correct RhD coding sequence (not shown).
As with MEL-C88 data, these results strongly suggested that human RhD protein would not be expressed in mouse erythroid cells in the absence of human RhAG despite the presence of murine Rhag.

Expression of hRhAG, but not hRh alone, in transgenic mice derived with pGSEL1 vectors
Since BAC injection did not produce mice transgenic for hRhAG, nor mice expressing RhD on red cells, transgenesis of hRHAG and RHD cDNA under control of the β-globin LCR/ promoter/enhancer was carried out using C57Bl/6 and B6/CBA mice. No founders were obtained on C57Bl/6 background. A transgenic line for hRHAG, TG_hRHAG_68.08, was obtained with B6/CBA ( Figure S1A, Southern blot). These mice were positive for erythrocyte membrane expression of hRhAG as confirmed by flow cytometry (Figure 2), with 45 to 75% positive erythrocytes in individual mice, and hRhAG sites about ten times lower than on human erythrocytes (i.e. 8000 as compared to 80 000 sites) (not shown). Human RhAG was evidenced by immunoblot on TG_hRHAG_68.08 erythrocyte membranes, not on TG_RHD_BAC1, or WT controls ( Figure  3). Glycosylation of hRhAG appeared more limited in mouse than human erythrocytes, where a large diffuse band from 30 to 70 kDa can be detected (Figure 3).
A transgenic line, TG_RHD_65.08, was also obtained for RHD on B6/CBA background ( Figure S1A). Again, no RhD was detected on the erythrocytes ( Figure S2), indicating that coexpression of RHD and hRHAG transgenes is required.

Mice co-expressing hRhAG with RhD under autologous or β-globin promoter
Transgenics TG_RHD_BAC1 and TG_RHD_65.08 were crossed with TG_hRHAG_68.08, and erythrocytes from the double transgenic mice (dTG_RHD_BAC1 and dTG_RHD_65.08, respectively) were analysed for expression of hRhAG and RhD. By immunoblot RBCs from these double transgenic mice both expressed hRhAG, but only dTG_RHD_65.08 showed a (low) level of RhD antigen expression ( Figure 3).
Flow cytometry (Figure 4) of RBCs from dTG_RHD_BAC1 and dTG_RHD_65.08 mice showed that both hRhAG and RhD were expressed, with only hRhAG-positive erythrocytes expressing the RhD antigen, a relationship noted in MEL cells (see Figure 1C). Similarly to single transgenic hRhAG mice, 40-70% of RBCs were hRhAG-positive (MFI 10-20% of value for human RBCs). RhD expression levels and the percentage of positive erythrocytes (2 -15% in individual mice) remained low in both double transgenic lines, as compared to the strong expression on control human erythrocytes ( Figure 4B, last dot plot). Interestingly, RhD in dTG_RHD_BAC1 was detected only from a high "cut-off" level of hRhAG expression, while in dTG_RHD_65.08 (under β-globin regulation), erythrocytes positive for RhD were observed for variable levels of hRhAG, (compare second and third plots Figure 4B).

Expression levels for human RHD and mouse Rh genes
Expression levels for mouse Rhd and human RHD were compared in single transgenic TG_RHD_BAC1 or double transgenic dTG_RHD_BAC1 mice, using mouse transferrin receptor1 as endogenous control ( Figure S3). Analysis was caried out on the BAC transgenic line as this allowed the use of an exon-spanning probe for RHD. We found that when the BAC1_RHD transgene was present, human RHD transcript expressed at a level similar to mouse Rhd transcript. Though human RhAG protein was necessary for RhD protein membrane expression, presence or absence of a human RHAG gene did not change the level of RHD transcript.

Interaction of hRhAG with mRh proteins on erythrocytes of mice transgenic for hRhAG
Co-Immunoprecipitation of mRh but not mRhag with hRhAG. Immunoprecipitation of hRhAG by LA18.18 was performed on ghosts of TG_hRHAG_68.08 erythrocytes, and immunoblots of immunoprecipitated material were probed with polyclonal antibodies reacting with mRh (MPC8), mRhag, or LA18.18 as an immunoprecipitation control. The control showed that hRhAG was correctly immunoprecipitated from red cell ghosts of TG_hRHAG_68.08 transgenic mice (IP control on Figure 5,left) and migrated as a diffuse band of the expected size.
LA18.18 immunoprecipitates from three transgenic TG_hRHAG_68.08 mice (TG1, 2 and 3) and a WT littermate were analyzed by SDS-PAGE and immunostained with crossreactive anti-Rh antibody MPC8. A strong signal at the expected size of Rh protein was detected in the three hRHAG transgenic mice, but not in WT ( Figure 5, middle). This band represents mRh protein, as there is no human Rh in these mice.
In contrast, when LA18.18 immunoprecipitates from TG_hRHAG_68.08 were immunostained with the antibody to mRhag, no signal could be detected ( Figure 5, right) indicating that mRhag did not co-precipitate with hRhAG. Controls for this last reaction are shown in Figure S4. Thus, only mRh proteins are detected in LA18.18 immunoprecipitates, suggesting the presence of mixed-species oligomers of hRhAG and mRh proteins, but not of hRhAG with mRhag.

Re-expression of mRh on erythrocytes of mRhag knockout mice transgenic for hRhAG
To examine whether mouse Rhag protein competes with human RhAG, expression of hRhAG in the absence of mRhag was investigated by crossing TG_hRHAG_65.08 mice with mRhag knockout (KO) mice [19]. Flow cytometry of RBCs from siblings of these crosses showed presence of hRhAG, but no increase of expression compared to WT/hRHAG mice: in one group of siblings the estimated number of hRhAG sites was 9534+2300 for mRhag -/-/hRHAG (n=4), 8567+2030 for mRhag +/-/hRHAG (n=4) and 9129 and 7648 for two mRhag +/+ /hRHAG. The number of sites remained approximately ten times less than for human erythrocytes (in this experiment 8x10 4 sites).

Ammonium transport in erythrocytes of mRhag -/-mice expressing hRhAG/mRh oligomers
To examine ammonium transport in erythrocytes expressing the heterologous RhAG complexes, pH-dependent fluorescent changes of resealed red cell ghosts from hRhAG transgenic mRhag KO mice (mRhag -/-/hRHAG) and their mRhag KO (mRhag -/-) controls, subjected to a 10 mEq inwardly directed ammonium gradient in iso-osmotic conditions, were measured with a stopped-flow instrument [28]. The alkalinisation rate constants (k) of ghosts, calculated from intracellular pHi elevation, were 1.45 ± 0.04 s -1 for mRhag KO and 1.71 ± 0.08 s -1 for hRhAG transgenic mRhag KO animals. The size of ghosts, as measured by light microscopy, was the same for the two genotypes respectively 4.73 ± 0.39 and 4.68 ± 0.42 µm. Thus, hRhAG transgenic mRhag KO erythrocytes showed a modest but significant (p<0.001) increment in ammonia transport when compared to mRhag KO. The apparent unitary permeability to NH 3 (P unitNH3 ) per hRhAG was 2.0 E-0.3 (Table  S2), similar to that measured in normal human erythrocyte ghosts (1.89 E-0.3 [28] and 2.19 E-0.3 calculated from data in  [29]), indicating that human hRhAG protein expressed in mouse retains its function, even when a fraction is associated with the mouse mRh protein.

Discussion
This is the first description of transgenic mouse lines expressing the human RhD protein. RhD could be observed only in mice transgenic for both RHAG and RHD human genes, consistent with the Rh complex model [7][8][9]32].
Expression of recombinant hRhAG, RhD and Rhc in the murine context was first examined using MEL-C88, a cell line previously used to express human blood group antigens, though not Rh [33]. In MEL cells hRhAG was expressed alone, but human Rh antigen (D and c) expression strictly required hRhAG, as expected from analysis of Rh null variants (reviewed in [8], expression studies using human K562 and HEK293 transfectants [27] and the study of mRhag -/-mice [19]. Expression of hRhAG/Rh proteins under control of the human β-globin LCR/ promoter/ enhancer did not significantly improve after erythroid differentiation, when compared to the EF1α promoter. MEL transfectants expressed relatively low levels of hRhAG and still lower levels of RhD or Rhc antigens, contrasting with expression levels 'similar to those in RBC' reported for Knops, Kell and Duffy antigens [33]. However, these antigens are expressed on red cells at much lower levels than RhAG/Rh antigens which have a complex membrane topology and whose assembly, traffic and stabilisation in the red cell membrane are currently undefined. Low expression levels had been observed when human K562 erythroleukemia cells, which express substantial amounts of endogenous hRhAG were forced to express human Rh antigens by retroviral-or plasmid-mediated cDNA transfer [27,34], even when co-expressed with recombinant Band 3, reported to enhance Rh antigen expression [35,36]. Of note, upon erythroid differentiation of MEL-C88/hRhAG cells by DMSO, murine band 3 was induced (from an initial 4% to 64% on day 6) and hRhAG expression was also enhanced, consistent with the existence of a macro-complex associating the Rh complex and Band 3 [37]. However, since RhD expression level was directly correlated to hRhAG level in both MEL-C88 ( Figure 1C) and K562 cell lines [27], it is postulated that hRhAG represents the essential limiting factor for Rh(D or c) protein expression.
In MEL cells the differentiation process might not have developed normally or gone to completion. In the transgenic mice, analysis of RBCs which had followed the entire murine erythroid differentiation programme in vivo was possible.
As in the cell lines, animals transgenic for the RHD gene alone (in the absence of hRHAG) did not express any RhD antigen, though transcript for human RHD was present, suggesting that endogenous mRhag is unable to assemble and/or transport human RhD protein to the red cell surface. In the hRHAG transgenic line, the mean level of hRhAG on RBCs, though highly heterogeneous, was about 10% of that on human erythrocytes. hRhAG could be co-immunoprecipitated with endogenous mRh, but not with mRhag, indicating a potent interaction between hRhAG and mRh. When hRHAG transgenic animals were developed on an mRhag -/- [19], as compared to a WT background, the level of hRhAG on their erythrocytes was not significantly increased. This suggests that the low levels of hRhAG are not simply a result of competition with mRhag for transport to the membrane, but might reflect differences in Rh complex processing. Very interestingly, immunostaining of RBC membrane proteins from the mRhag -/mice transgenic for hRhAG revealed that mRh, absent from the mRhag null RBC membranes, was re-expressed when hRhAG protein was present. This result was consistent with the formation and trafficking to the membrane of a mixed-species oligomer complex between hRhAG and mRh.
From these cellular expression models and single transgenic mice, it appears that hRhAG-mRh and hRhAG-RhD oligomers are present in the mouse cells in addition to endogenous mRhag-mRh oligomers, but not associations of mRhag-RhD or mRhag-hRhAG. We analyzed the interactions between human and mouse Rh/RhAG proteins in light of 3D structure of human RhCG, solved at 2.1 Å resolution (pdb 3HD6) [12] to build accurate models for the human RhAG/RhD and mouse Rhag/Rh subunits, within the context of a trimeric architecture. These models are refined relative to those we previously published, based on the bacterial ammonia transporter AmtB [38]. Information collected from careful examination of interfaces between RhAG and RhD subunits, within the human RhAG/RhD and mouse Rhag/Rh heterotrimers, is summarized in Figures S5,S6,S7. The main observation deduced from this analysis is that the trimeric interface includes several aromatic amino acids, some of which are well conserved between the subunits and between the human and mouse sequences. Striking differences between the human and mouse sequences concentrate in two symmetrical external areas of the interfaces between the RhAG and Rh subunits ( Figures S5,S6,S7). This is certainly not the only factor to be considered, due to the large surface involved in the trimer assembly, but the region highlighted here is likely to be one of those which may play a critical role.
When the mice were transgenic for both hRHAG and RHD on a WT background, the two human proteins, hRhAG and RhD, were expressed on RBCs, as opposed to the absence of RhD expression in the RHD single transgenics. The human RhD polypeptide expressed was recognized by polyclonal anti-D Rhophylac as well as by monoclonal anti-Ds reacting with different epitopes, respectively Ep 3.1 (LOR15C9) Ep6.2 (H2D5D2F5) and Ep13.1(LFB-R593) [30], and thus would appear to be assembled in the membrane in identical fashion to that in human red cells.
When compared to MEL cells, expression of the two proteins was less directly correlated, and hRhAG appeared to be in excess of RhD antigen. This suggests that hRhAG might not be the only limiting factor for RhD antigen expression, even taking into account a possible competition of mRh with RhD. A future analysis of double transgenics on an mRhag -/-mRh -/background may clarify this issue. Thus, although hRhAG is essential for RhD expression, it is not sufficient to promote full membrane expression of Rh antigens at the cell surface.
Factors that may limit heterologous hRhAG/Rh expression in transgenic animals at the transcription level include insertion site and transgene copy number, not studied here, as only one transgenic line could be established in each case. The difference in expression pattern of hRhAG/RhD under BAC or β-globin regulation might be linked to differing nuclear organization of transcription for the promoters (reviewed in Cope et al. [39]). Transgene expression outside of the original gene locus is often anarchic, either very high or very low, and may be prone to variegation [40]. However, when RHD transcript expression was studied in the BAC transgenic lines, we found that found that whenever the BAC1_RHD transgene was present, human RHD transcript was expressed, and at a level comparable to that of mouse Rh. We also found that presence or absence of a human RHAG gene did not impact the level of RHD transcript. These observations are in favor of a post-transcriptional nature of the limitations in RhD expression.
A number of factors may play a role in intracellular traffic and/or cell surface expression of RhD and RhAG on mouse RBCs. Impairment of N-glycosylation has been shown to disrupt the targeting and stability of membrane protein expression, notably in the case of cytokine receptors, and transporters (reviews in [41,42]), though conservation of core glycosylation was sufficient for correct cell surface expression of the cystic fibrosis transmembrane conductance regulator [43]. Posttranslational processing of human RhAG in mouse and human erythroid cells differs, our results show that hRhAG glycosylation is reduced in the mouse, bands are less diffuse, and always below 50kDa on SDS-PAGE. How altered glycosylation of hRhAG might affect trafficking mechanisms for efficient cell surface expression of Rh antigens, or be a reflection of trafficking defects, is currently not clear.
Differences in regulation of trafficking of RhAG to the cell membrane in man and mouse have been discussed recently [44], with emphasis on the putative role of protein 4.2 and GPB (glycophorin B), both present in the Rh macrocomplex. In man, in the absence of protein 4.2 there is increased glycosylation of RhAG [20,21], this is also seen in GPB-deficient RBCs [45]. In contrast, in the mouse, endogenous mRhag glycosylation is not affected by protein 4.2 deficiency [20,21]. However, in both cases (man or mouse), expression of Rh antigens is normal in 4.2-deficiency [20,21].
The lack, or inadequacy of a mouse protein partner for transit or membrane stabilisation could also affect hRhAG trafficking. In mice the Gypb gene is absent [46], but this accessory protein is also dispensable in the human Rh complex, since GPB-deficient individuals express Rh blood group antigens normally [45]. CD47 is independent of the Rh complex in mouse. As concerns ICAM4, erythrocytes of human LWa-/ LWb-individuals, lacking ICAM4, are not deficient in Rh antigens [7], and we found no difference in ICAM4 between transgenic and WT mice (data not shown). Differences implicating these known accessory Rh complex proteins should thus not significantly compromise cell surface expression of the exogenous hRhAG/RhD protein core.
Recent studies have shown that the Rh complex is an important interaction site between the lipid bilayer and the spectrin-based membrane skeleton, that regulate the shape, deformability and mechanical properties of RBCs. Data from protein 4.1R-deficient mice led to the proposal that a proportion of mRh, interacting with the 4.1R-based junctional complex [47], was independent of mRhag. However, in our mRhag-/knockout and hRhAG transgenic mice, we could detect no fraction of independently bound Rh, and mRh was never found at the red cell membrane without either mRhag or hRhAG. These observations tend not to favor an impact of protein 4.1R on RhD expression, all the more so as, in human protein 4.1null red cells, the Rh protein was not altered (unpublished data referred to by [17]). The critical role of ankyrin 1 in Rhmembrane skeleton linkage and the relationship with the Band 3 macrocomplex have also been underlined (reviewed in [17,44]). Ankyrin 1 is an essential partner, anchoring RhAG and Rh proteins at the red cell surface. mRhag and mRh are severely reduced on RBCs from ankyrin1-deficient nb/nb mice [48], and (Rh) in ENU Ank1 null mice [49]. One mechanism for deficiencies of mRhag/mRh in nb/nb mice appears to be abnormal sorting during erythroblast enucleation [50]. Whether mouse ankyrin 1 can interact with human RhAG is not known. Homology of subdomain 2 of membrane binding domain in mouse and human ankyrin 1 [48] would be in favor of this, as well as the observation of hRhAG at the mouse RBC membrane.
No clear differences involving identified partners of the Rh complex appear to explain the reduced efficiency of hRhAG and RhD expression in the mouse RBC -as yet unknown elements in Rh complex trafficking and stability may be implicated.
Our studies demonstrate that it is possible to express the human RhD antigen on erythrocytes in mice, provided the human Rh glycoprotein (hRhAG) is co-expressed, as mouse Rhag appears unable to direct RhD protein to the cell membrane. The development of mice expressing human RhAG and RhD represents a novel approach to dissect Rh/RhAG complex formation, in complement to studies of human erythroid differentiation in vitro [51]. The production of transgenic animals for human Rh antigens is a step towards the development of humanized Rh-positive and Rh-negative mouse lines. These should prove useful in modelling the immune response to Rh antigens -allowing analysis of Rh immunisation and mechanisms of antibody-mediated immune suppression in mice whose immune repertoire has developed in the presence of the human Rh proteins. This model could provide a basis for the evalution of the protective mechanisms of anti-Rh D prophylaxis, and an experimental basis for the evaluation of monoclonal antibody efficacy.  Flow cytometry analysis shows that erythrocytes from hRHAG-RHD double transgenic mice (dTG_RHD_BAC1 or dTG_RHD_65.08) obtained by crossing TG_RHD_BAC1 or TG_RHD_65.08 with TG_RHAG_68.08, respectively, express RhD antigen, while erythrocytes from RHD single transgenics (TG_BAC1_RHD or TG 65.08_RHD) do not. Red cells labeled for IgG (control) and RhD expression (anti-D LOR15C9). (PDF) Figure S3. mRh or hRHD expression normalized to transferrin receptor expression in RHD and in RHD/RHAG transgenic mice. Transcript expression was measured in single (BAC1_RHD) transgenic mice and in mice transgenic for human RHAG and RHD (dTG BAC1_RHD)-four mice of each type. In BAC1_RHD transgenic mice, human RhD transcript was expressed at a level comparable to that of mouse Rh. In double transgenic mice, the presence of the human RHAG gene did not increase the level of RHD transcript as compared to the single transgenics (Student's unpaired t-test p=0.45). (TIF)  Figure S5. Sequence alignments. Alignment of the sequence of the human RhCG subunit (whose 3D structure has been solved, pdb 3HD6) [12] with the sequences of hRhAG, mRhag and RhD. The human RhCG sequence shares 51 % and 32 % with the hRhAG and RhD sequences, respectively (49 % and 32 % with the mRhag and mRh sequences). The observed secondary structures are shown above the alignment. Squares indicate positions participating in the interface between the subunits of a trimeric assembly, as deduced from the observed significant differences in solvent accessibility between the monomeric and trimeric subunits. Green squares indicate amino acids that are identical between hRhAG and RhD and between the mouse and human sequences, orange those which are different between hRhAG and RhD but identical between the human and mouse sequences, red those which differ between hRhAG and RhD and between the human and mouse sequences. (PDF) Figure S6. Analysis of 3D structures.Model of the 3D structure of the human RhAG/RhD(2) heterotrimer, based on the experimental 3D structure of the human RhCG homotrimer (pdb 3HD6) [12] and the alignment given in Figure S4. The amino acids participating in the heterotrimer interface are shown and colored according to Figure S4). These involve residues from M0 in one subunit and residues from M7 and M9 in the other subunit. In particular, three amino acids from M7 (F245 in human RhAG, Y243 in human RhD) and M9 (H292, P293 and F294 in human RhAG, S290, P291 and W292 in human RhD) seem to play a critical role in the interface formed with helix M0 (involving human RhD F28 and Y29 and human RhAG F20 and G21). The large substitutions observed with other amino acids in the equivalent mouse sequences in the area of the human RhD M0/RhAG M7-M9 interface (human RhAG F245 with M255, human RhAG H292 with P302, F294 with Y304, human RhD Y29 with C29, H33 with P33 and Y34 with H34) may lead to substantial differences, which may preclude the formation of a stable heterotrimer mRhag/RhD(2) chimera. (PDF) Figure S7.

Supporting Information
Differences between human and mouse sequences. Magnification of two areas from Figure S5 concentrating major differences between human and mouse sequences. (PDF) input concerning stopped-flow experiments, S. Bigot for flow cytometry experiments and ghost size determination, and C Lopez for reading the manuscript . In memory of Charles Babinet who advised and encouraged this project.